There exists a need for a highly sensitive and specific technology directed to the detection of human pathogens and toxins in food, water, and the environment. It is very difficult to effectively detect organisms in natural fluids such as milk, blood, sewage and meat products at low concentration and to discriminate between pathogenic and harmless species. Conventional bioassay methods are commonly designed for samples on the order of a few cubic centimeters, and the extraction or concentration of pathogenic material from larger volumes to met sensitivity requirements creates additional challenges.
One of the most promising strategies for performing pathogen assays on raw, unpurified samples is based on sensors that harness biological ligand-receptor interactions to identify specific compounds. Examples of approaches that implement such a strategy include fiber optic evanescent wave sensors and surface plasmon resonance sensors.
An electromagnetic wave, traveling through one material, that is reflected at a dielectric interface produces an exponentially decaying electric field within the second material on the opposite side of the interface. At optical frequencies this is termed the evanescent wave effect, and at radio frequencies this phenomenon is often called a “skin effect.” The penetration depth within the second material, the evanescent wave region, is a small fraction of a wavelength, yet greater in size than most optical labels 100 such as light- or fluorescence-producing reporter molecules, light-absorbing or scattering molecules, and colloidal particles and microspheres. These labels can be used to monitor or produce optical changes in the evanescent region, or modify the propagation of light in the adjacent dielectric, providing a fundamental means of detecting target materials that are close to the surface while discriminating against those far away. In particular, by coating the interface with a capture agent that is specific for a microscopic or molecular target of interest, exquisitely sensitive optical-based sensors can be created.
In one competitive assay technique, fluorophore-labeled antigen 104, together with the sample to be tested, is exposed to the coating of capture antibody on the fiber, and the labeled antigen competes for antibody binding sites with non-tagged analyte 106 in the test sample. The evanescent field produced by light 108 passing through the fiber 102 then excites the fluorophores into light emission 110, and the fiber itself conveniently acts as a return waveguide for the fluorescent signal. In this example, the strength of the fluorescent signal is inversely related to the analyte concentration in the test sample. Alternatively, a non-competitive technique, such as a sandwich-format assay, can be used, in which case the fluorescent signal is directly related to the analyte concentration in the test sample. High sensitivity and specificity can be achieved for a wide range of metals, toxins, proteins, viruses, living and dead bacteria, and spores, through the use of bound target-specific agents 100 such as chelating agents, antibodies, crown ethers and the like, combined with appropriate optical labels that luminesce, fluoresce or alter light transport by the waveguide. In applications where pathogens will be infrequently found, cost per assay may be low since the sensor remains active until the capture agents have been substantially neutralized by-the binding of the target material.
For surface plasmon resonance sensing, FIG. 1B shows a thin layer of metal 110, such as gold, applied to a core portion 112 of an optical fiber 114 from which the cladding 116 of the fiber has been partly removed. The evanescent electric field produced by light 118 passing through the fiber 114 excites surface plasmon waves 120 on the outer surface of the metal 110. When white light is passed through the fiber 114, the excitation of a surface plasmon wave causes a dip in the spectrum of the light passing through the fiber, with the dip occurring at a resonance wavelength which is a function of the complex indices of refraction of the fiber core, the metal layer, and the solution surrounding the fiber, as well as the incidence angle of the light. Light passing through the fiber 114 can be returned by a mirror 122, or can be passed through the distal end of the fiber (in the absence of a mirror) for optical processing and analysis, as is well known to those skilled in the art. Any change in the index of refraction of the solution is detectable, and molecules binding to the surface of the metal 110 can then be detected if they have an index of refraction that is different from the bulk solution. Coating the metal layer 110 with target-specific capture molecules (not shown), which react with target analytes within a sample solution, then allows detection of reactions (such as antigen-antibody reactions and reduction-oxidation reactions) on the surface of the metal.
Fiber optic evanescent wave sensors are the subject of a number of U.S. patents, including the following, the disclosures of each being incorporated herein by reference: U.S. Pat. No. 4,447,546, to Hirschfeld et al., entitled “Fluorescent Immunoassay Employing Optical Fiber in Capillary Tube”; U.S. Pat. No. 4,558,014, to Hirschfeld et al., entitled “Assay Apparatus and Method”; U.S. Pat. No. 4,582,809, to Block et al., entitled “Apparatus Including Optical Fiber for Fluorescence Immunoassay”; U.S. Pat. No. 4,654,532, to Hirschfeld, entitled “Apparatus for Improving the Numerical Aperture at the Input of a Fiber Optic Devices”; U.S. Pat. No. 4,716,121, to Block et al., entitled “Fluorescent Assays, Including Immunoassays, with Feature of Flowing Sample”; U.S. Pat. No. 4,909,990, to Block et al., entitled “Immunoassay Apparatus”; U.S. Pat. No. 5,242,797, to Hirschfeld, entitled “Nucleic Acid Assay Method”; U.S. Pat. No. 5,061,857, to Thompson et al., entitled “Waveguide-Binding Sensor for Use With Assays”; U.S. Pat. No. 5,430,813, Anderson et al., entitled “Mode-Matched, Combination Taper Fiber Optic Probe”; U.S. Pat. No. 5,152,962, to Lackie, entitled “Immunoassay Apparatus”; U.S. Pat. No. 5,290,398, to Feldman et al., entitled “Synthesis of Tapers for Fiber Optic Sensors”; and U.S. Pat. No. 5,399,866, to Feldman et al., entitled “Optical System for Detection of Signal in Fluorescent Immunoassay.” Fiber optic surface plasmon resonance sensors are the subject of U.S. Pat. No. 5,359,681 to Jorgenson et al., entitled “Fiber Optic Sensor and Methods and Apparatus Relating Thereto,” the disclosure of which is incorporated herein by reference.
For evanescent wave sensors, it is desirable to optimize the magnitude of the evanescent electric field as well as to optimize the optical properties of the return path for the detected fluorescence. The above-identified patents describe numerous optimization approaches, including attempts to match the numerical aperture of various system components and to improve system numerical-aperture. Numerical aperture is a measure of the largest angle, relative to the optical axis of a system, that a ray of light can have and still pass through the system. Each component in an optical system will have its own unique limiting numerical aperture, and the maximum system numerical aperture will be determined by the system component having the lowest numerical aperture. The system numerical aperture is a key parameter in optical sensing since transferred power is typically proportional to its square. Good design practice and cost efficiencies require system components to have matching numerical apertures.
One well-known approach of matching numerical apertures employs tapered or cone-shaped waveguides. In addition to providing numerical aperture matching, tapering the active, analyte-sensitive portion of the optical fiber maintains a substantial fraction of the input light near the critical angle, thereby maintaining a high magnitude evanescent field. However, there is also a constant loss of light along the sensor fiber as the taper acts upon rays that are already only weakly guided and causes them to exceed the critical angle.
In order for white light to propagate in an optical fiber used in connection with a surface plasmon resonance sensor, the fiber must have a large enough diameter to support the longest wavelength of light. Also, a large diameter fiber propagates higher numerical aperture light, which makes it easier to excite surface plasmon waves in metal films of a thickness readily fabricated by conventional processes. As a consequence, multi-mode fibers are used which propagate light over a range of angles. However, this range of angles results in a less distinct resonance effect, because each angle of propagation results in a different resonance wavelength.
FIG. 2A shows the theoretical resonance curves for various propagation angles relative to the optical axis of the fiber core, assuming a 55 nm thick layer of gold on a silica optical fiber core immersed in water. The overall resonance detected is a superposition of the resonance effects for each of the various angles of propagation. FIG. 2B shows the integration of individual theoretical resonance curves for propagation angles from 0 to 23.6 degrees, assuming a sine-squared distribution of optical; power at the various propagation angles. The significant signal degradation associated with current approaches to surface plasmon resonance sensing is seen by comparing the resonance curve of FIG. 2B with the individual resonance curve of, for example, 23.6 degrees in FIG. 2A.
The first evanescent waveguide sensors, described in the early 1980's, were for substantially cylindrical waveguides, that is, waveguides with circular cross-sections in which light uniformly filled the entire cross-sectional area. Recent development has strongly emphasized planar waveguides excited by collimated light beams. In these devices, light is only contained in one dimension and lateral spreading is totally defined by excitation optics. This substantial shift has occurred primarily due to an interest in creating multianalyte assay arrays by printing a linear or two-dimensional pattern of capture agent spots on one surface of the planar waveguide within the illumination path of the light beam, and then monitoring for an optical signal from individual analyte-specific spots with a CCD detector array or photomultiplier on the other side of the slab waveguide.
However, the planar approach has some other weaknesses in addition to its limited light guiding ability. Due to the typically small size of individual assay spots it is a challenge to effectively contact each dot with the entire fluid sample. This is of particular significance when foodstuffs are tested for pathogens. Regulations may require, because of high health risks at extremely low pathogen levels, that assay samples of 300 cubic centimeters or more be utilized. By way of example, the acceptance limit set by the US Department of Agriculture for Escherichia coil O157:H7 is one organism per 25 gm of sample. It is very difficult to effectively detect organisms at such a low concentration with methods based on bioassay dots of typically 1 mm^2 or less area. In addition, sample heterogeneity becomes an issue when raw food samples are examined. Fat globules and other non-toxic components may adhere non-specifically to the sensor or physically block contact with the target, reducing the effective sensitivity. Samples may also be viscous which increases the mass transfer boundary layer thickness and decreases the diffusive mass transport rates. These factors may yield low signal levels and create poor assay statistics where the target is a low, yet lethal concentration of a human or animal pathogen.
Analyte mass-diffusion boundary layers are also typically thicker for planar structures than for solids of revolution, such as cylinders. For related reasons a planar geometry may be more difficult to clean if the assay involves a multi-step protocol such as a sandwich immunoassay, or if it is desired to reuse the sensor. Finally, for applications such as food safety the number of target pathogens may be only one to six, calling into question the value of low sensitivity array techniques that require sophisticated and possibly costly CCD or photomultiplier signal recovery techniques.
Although evanescent wave and surface plasmon resonance sensors show great promise for use in medical and food safety applications, those skilled in the art understand that the current technology is less than optimal in a number of respects, including those disadvantages identified above.